Using fuzzy logic for on-line trend analysis

Real-time trend analysis of plant data is an important but sometimes difficult human-thought process. While experienced operators can usually quickly filter out spurious information upsets, novices must spend considerable time acquiring such skill. Then the number of trend analyses are high, updating the memory of current plant situations can be onerous, resulting in omissions or other failures. A real-time expert system has been developed to assist in this task for a copper flotation process. Using fuzzy logic, the system determines current and past process states and reports on DCS signals regarding significant changes up or down. When such a change occurs, a message is sent to the computer console to indicate the trend and its corresponding time scale. The system filters out faulty assays and recognizes when equipment is shut down. In this way, an operator can instantly become aware of process trends upon returning to his station from an excursion into the plant. Aspects of the trend analyzer are discussed in this paper and an example of its successful application is given.<>